Skip to main content
crunching numbers

Whenever a party is confronted with a yawning gap between themselves and their main rivals in a pre-election poll, it is usually waved off because "campaigns matter." And they do. Campaigns give parties the opportunity to present a platform to Canadians and give voters a chance to get to know the leaders and their local candidates better. Campaigns also tend to shift pre-election polling results a great deal.

An analysis of the last 30 years of federal elections indicates that in almost every case the pre-election voting intentions of Canadians shift significantly by the time the vote is held. But the degree to which voting intentions change can vary just as significantly.

If there is to be an election this spring, it's likely to be held in early May. According to polls conducted this month, the average support for the Conservatives stands at 37.1 per cent, while the Liberals are at 26.1 per cent and the New Democrats are at 16.6 per cent. Looking back at polls taken three months before each election since 1980 tells us we should expect these numbers to change.

The variation between the average results of polls taken three months before an election over the last 30 years is +/- 4.8 points per party. In most cases, the variation is between 1.3 and 3.7 points, but the elections held between 1984 and 1993 featured much wilder swings.

In June 1984, John Turner's Liberals stood at 45.5 per cent only to find themselves bottoming out at 28 per cent in September. This shift benefited both Brian Mulroney's Progressive Conservatives, who went from 41 to 50 per cent, and Ed Broadbent's New Democrats, who saw their vote increase from 12 to 18.8 per cent.

In 1988, the Progressive Conservatives jumped again, gaining 7.5 points and ending up with 43 per cent of the vote on election night. This time, it was the New Democrats who dropped most. In August 1988 the party had the support of about 27.5 per cent of Canadians. Three months later, the NDP took in 20.4 per cent of the vote.

But 1993 is probably the best example of how campaigns can change pre-election polling results. In August 1993, Kim Campbell's Progressive Conservatives and Jean Chrétien's Liberals were statistically tied at 35 and 36.5 per cent support, respectively. But on election night, the PCs lost 19 points and were reduced to only 16 per cent support, going from 169 seats to only two. The Liberals picked up about five points, as did the New Democrats, but it was the Reform Party under Preston Manning that changed the game, increasing their support from 9 per cent in August to 18.7 per cent at the end of October.

However, not every election features these kinds of dramtic shifts. Between November 1979 and February 1980, Pierre Trudeau's Liberals lost about four points to Joe Clark's Progressive Conservatives, but that did not get in the way of Mr. Trudeau forming a majority government after that election, as the polls had predicted he would three months earlier.

In 1997 and 2000, the Liberals dropped 9.5 and 7.4 points, respectively, in the three months preceding the elections held in those years. But in each case the Liberals still won a majority of seats.

More recently, the 2004 election saw very little change in the last three months before the vote. In March 2004, the Liberals stood at 37.3 per cent support, ahead of the Conservatives at 27.7 per cent and the New Democrats at 17.7 per cent. In the June election, the Liberals ended up with 36.7 per cent of the vote, compared to 29.6 per cent for the Tories and 15.7 per cent for the NDP.

When Stephen Harper won his first national election, there was a major shift between October 2005 and January 2006, with the Conservatives gaining almost nine points, most of which came from the Liberals. But the New Democrats and even the Greens stood relatively still.

And in 2008, the Conservatives widened their lead from the 34.3 to 30.5 per cent margin that existed in July to the 37.7 to 26.3 per cent gap on voting day.

History is not on Liberal Leader Michael Ignatieff's side. In the nine campaigns since 1980, the Liberals gained support over their pre-election polling levels only once. With the exception of the debacle in 1993, the Conservatives and their predecessors have gained every time.

If there is a silver lining for the Liberals, it is that in six of the nine elections it was the government that lost support. On average, the governing party has lost 5.1 points during the three months preceding an election since 1980, while the main opposition party has gained 2.7 points.

On average the Conservatives and their predecessors have gained 3.3 points in the three months preceding an election. The Liberals, on the other hand, lose an average of 5.3 points.

If we are to expect a shift in support between now and a potential election in May, it will almost certainly be a shift between the Liberals and the Conservatives. On average the New Democrats have only lost 0.3 points over the three months preceding each election, while their overall change in vote is as low as +/- 2.8 points. The Liberals and the Conservatives, on the other hand, can expect their vote to shift by as much as +/- 6.5 points by May.

But if there is one thing to take from the history of pre-election polling, it is that change does not come easily. In only two of the last nine elections has the lead shifted from one party to another, meaning the next election is Stephen Harper's to lose.

Éric Grenier writes about politics and polls at ThreeHundredEight.com

Interact with The Globe